One main characteristic of the discourses that express our beliefs, including hypothetical discourses in science, is that they serve as guides for action. This pragmatist dimension is present in Plato. An epistemological consequence of this largely neglected aspect of his philosophy is that these discourses cannot be directly taken as true. Plato always insisted that he did not know the truth regarding the issues he critically discussed with his interlocutors. Nevertheless, those who affirm that the best critically established discourse is wrong, without providing a good reason, are talking nonsense. Plato also establishes the bases for any form of communicative rationality and the issues that are involved in rational theory change. These include revision of the presuppositions and methodology that we use to criticise our previous beliefs and establish new ones, and the difficulties of undertaking this latter task. I analyse the two possible scenarios that emerge when we critically revise our beliefs. In the first scenario, we can detect that our previous beliefs were wrong without being able to establish a better alternative, being prey to what I term ‘doxastic dizziness’ as a consequence. The second scenario is that we end up agreeing on a discourse that we deem rationally better than the previous one without any meta-epistemological guarantee that it is true. It is, nevertheless, more epistemically verisimilar.